Voyage

Inspiration

In a world where 80% of Gen Z in the U.S. turns to social media for travel ideas, the way we plan our excursions has taken a scenic route. With every swipe and tap on TikTok, Instagram Reels, and YouTube Shorts, we're not just passing time; we're curating our next big adventure, building itineraries off the well-documented experiences of others.

Interestingly enough, social platforms such as TikTok truly know the preferences of individuals better than the individuals themselves; we find this to be true with our own “for-you” pages, which somehow provide better recommendations for our upcoming trips than we can brainstorm ourselves (be it spring break, summer travel, or more)! However, the process through which we need to parse the media available on social sites to plan travel itineraries is extremely laborious and requires the manual input of different hashtags/keywords, and sorting through the outputs ourselves to choose what we’d be interested in. What if there was a way for information about our preferences to be utilized to build our dream trip?

Moreover, Gen Z is obsessed with learning more insights about themselves. Be it Marriage Pact, Spotify Blend, or other services that take user data and extrapolate fun insights, there exists a real need for breaking down data that exists and forming connections. What if there was a way to blend our own travel interests based on our social media preferences with those of friends we’ll be traveling with?

Introducing Voyage!

What it does

We built Voyage, a trip planner that distills a user’s “vibe” based on their TikTok preferences, subsequently generating activity suggestions for a given location presented in the form of 3 TikTok videos. We were able to achieve an algorithm to examine a TikTok user’s liked videos, sort their top hashtags, and generate the user’s vibe in three key adjectives (i.e. sporty, adventurous, curious, foodie, etc.). These adjectives are fed into GPT along with the location entered by the user to generate 3 topical activities at the destination, linking TikTok videos that depict the experiences.

How we built it

Creating Voyage was an adventure in itself, filled with learning curves and breakthrough moments. To achieve this task, we first utilized the TikAPI to retrieve the top 10 hashtags from a user's top 30 liked videos. Then, we fed these hashtags into the GPT-3.5-turbo model via the OpenAI GPT-3 API. The model generated three adjectives that described the user based on these hashtags.

Next, we fed these adjectives back to GPT to generate three activities that the user should do in a particular city they specified. Once the user specified the city, we integrated a search function to look up these activities on TikTok using the TikAPI. Finally, we displayed the TikToks related to these activities for the user to explore and enjoy. This process ensured personalized recommendations based on the user's interests and preferences.

Challenges we ran into

Challenges we ran into included learning how to use the TikTok API and deciding how to make sense of a user's profile and combine that into a person's 'vibe'. Initially, we attempted to use users' posts to pull a “vibe”, but quickly realized that most users don't post but rather save or like videos. Therefore, we pivoted to using the hashtags on their liked videos as data instead. However, some hashtags from multiple videos were repetitive, while others were irrelevant, such as “fyp” (for you page) or “ad.” To address this, we stored hashtags non-repetitively and ranked them according to the frequency at which they appear on the user's profile. Additionally, we filtered out irrelevant hashtags like “fyp,” which stands for the For You page.

Accomplishments that we're proud of

We're proud of several accomplishments throughout this project. Recognizing the common dilemma of having numerous saved travel inspiration TikToks that are never revisited, we devised a solution to leverage TikTok for trip planning more engagingly. Our approach aimed to help users discover more about themselves through their TikTok profiles while creatively organizing their travel plans. Throughout the process, we pivoted multiple times, brainstorming various methods to enhance the experience and ensure interactivity. This flexibility and continuous iteration were so valuable, and taught us so much! We were also able to build a super neat web portal to display our trip results.

The results we produced for our team members, Ishita and Arshia, were particularly gratifying; Arshia had actually already completed two out of three recommended activities our algorithm suggested for her specific to Paris!

What we learned

Reflecting on our journey, we've learned some valuable lessons. Firstly, we've learned that keeping things simple is key. While we started with big dreams and loads of ideas, we soon realized the importance of staying grounded and focusing on what we can actually achieve with the time and resources we have, while keeping the vision in mind. We also figured out the importance of boiling down to the MVP that people would really use, stripping away the added “nice-to-have” features.

Communication and task-sharing were big learnings too. We found out that dividing tasks based on our strengths and working in pairs on similar stuff, then coming back together regularly, was the most effective way to get things done. Anabelle and Ira tackled the algorithm and web portal, while Arshia and Ishita took charge of the customer experience design, and we found this approach to be very efficient!

Lastly, we learned to look at what people are already using and pay attention to what bugs them about those tools – it's where the best ideas for improvement often come from.

What's next for Voyage

Our next Voyage (no pun intended) involves implementing the blend feature, which generates a blended result for two individuals planning a trip together, catering to friends who wish to embark on adventures collectively. Once accomplished, our focus will shift towards creating a blend feature for groups, facilitating seamless group trips. Furthermore, we aim to extend our platform's reach by integrating with other social media channels such as Facebook, Instagram Reels, and YouTube Shorts, broadening our user base and enhancing the trip planning experience across multiple platforms.

Built With

• openai
• react
• tikapi